§3. Components

These are the building blocks of an extended mind. Each component exists at the intersection of memory and action — it is both something the system remembers and something agents act upon.

Component	What it is	Memory dimension	Action dimension
Protocols	Sets of instructions that govern how agents operate within the system. The rules of engagement.	The system remembers its own operating procedures. Protocols persist across sessions and agents.	Agents read and follow protocols before doing anything. Protocols shape every action.
Projects	Bounded units of work with a defined purpose, scope, and lifecycle. A project has a beginning, an arc, and (sometimes) an end.	The system remembers what each project is, its current state, its history, and its decisions.	Agents execute work within project boundaries. Projects are where ideas become artifacts.
Activity Log	A chronological record of modifications to the extended mind system. What was done, by whom, when.	The log is the raw material of long-term memory. Without it, the system cannot learn from its own activity.	Logging is itself an action — the final step of any work unit. It closes the feedback loop.
Skills	Reusable, structured operations that an agent can invoke. A skill has defined inputs, outputs, and scope.	Skills encode methodology. They are remembered procedures that standardize recurring operations.	Invoking a skill is a specific kind of action — one that guarantees a consistent outcome.
Tasks	Discrete units of work within a project. A task has a clear completion condition.	Tasks represent the system’s understanding of what needs to be done — working memory for execution.	An agent picks up a task, executes it, and marks it done. Tasks are the atomic unit of action.
Checkpoints	Milestones that mark significant progress within a project. What was delivered and what changed.	Checkpoints are compressed memory — they summarize phases of work into reference points.	Reaching a checkpoint is a meaningful action that resets context and informs next steps.
Tickets	Specific issues, bugs, or requests that need attention. A ticket has context, priority, and action steps.	Tickets are the system’s memory of problems encountered and requests received.	Resolving a ticket is an action. Tickets often originate from action (a bug found during execution).
Identity	An emergent organization of memory that represents who the operator is, how they work, and what they value.	Identity is meta-memory — memory about memory. It shapes how all other components are interpreted.	Identity informs action indirectly: it sets preferences, priorities, and tone. It is the “why” behind decisions.
Agents	The entities (H, NH, HNH) that perceive, remember, and act within the system. See §2.	The system remembers which agents exist, their capabilities, and their history of interactions.	Agents are the source of all action. Without agents, the system is inert.

Consolidation

Consolidation is the process of converting raw activity into structured knowledge — the agent equivalent of sleep consolidation in humans.

Activity Log (raw events)
  → Pattern Extraction (what recurs, what changed)
    → Long-term Memory (structured knowledge)
      → Identity (emergent model of the operator)

This pipeline is what separates storage from extension. A system that only logs activity is at Level 1–2 on the spectrum. A system that extracts patterns and builds identity from its own activity approaches Level 4.

Memory diagram

Strong (implemented)

Emerging (incomplete)

Planned